基于混淆熵的分类器集成方法  

Ensemble classifier method based on confusion entropy

作　　者：陈浩 任晓莹 李彦宏 薛佳炜 王淑琴[1] 崔玥 CHEN Hao;REN Xiaoying;LI Yanhong;XUE Jiawei;WANG Shuqin;CUI Yue(College of Computer and Information Engineering,Tianjin Normal University,Tianjin 300387,China)

机构地区：[1]天津师范大学计算机与信息工程学院,天津300387

出　　处：《天津师范大学学报（自然科学版）》2023年第3期69-73,共5页Journal of Tianjin Normal University：Natural Science Edition

基　　金：国家自然科学基金资助项目(61070089);天津市应用基础与前沿技术研究计划重点资助项目(15JCYBJC4600);天津市科技计划资助项目(19JCZDJC35100);天津市科技计划项目技术创新引导专项基金资助项目(22YDTPJC00610)。

摘　　要：本文提出了一种基于混淆熵(confusion entropy,CEN)的分类器集成算法.该算法按照候选分类器的CEN值升序排序,遍历选择使得CEN值减小的分类器,从而使组合不断优化,以获得更好的集成分类结果.使用7个候选分类器在5个UCI数据集上进行实验,结果表明,所提算法整体上优于经典的分类器集成算法AdaBoost和XGBoost以及另外3种近期算法(AdaCost、AdaCost-CNN和CU-AdaCost),验证了算法的可行性.A ensemble classifier algorithm based on confusion entropy(CEN)is proposed.The candidate classifiers are sorted in ascending order according to their CEN values,and then the algorithm sequentially traverses and selects the classifiers that reduce the CEN values,so that the combination can be continuously optimized to obtain better ensemble classification results.Seven candidate classifiers are used to perform experiments on five UCI datasets.The results show that the proposed algorithm is superior to the classical ensemble classifier algorithms AdaBoost and XGBoost,as well as other three recent algo-rithms including AdaCost,AdaCost-CNN and CU-AdaCost,which verifies the feasibility of the proposed algorithm.

关 键 词：分类器集成 集成学习 混淆矩阵 混淆熵 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]